I'm sure for some workloads, cassandra is 'ideal' or at least was 10 years ago.
What I can say though (as someone running thousands of cpus worth of cassandra at the moment); is that I deeply dislike this database. It's an operational nightmare, and I will not miss it even the slightest bit if it goes away forever.
On the other hand, I'm sure glad to see a bit of life in cassandra dev, it seems to have been dormant for a while; but would I pick it for new applications in 2026? No. I'd strongly recommend against using it unless you have the money to pay for a dedicated team to maintain it, and even then just.. No.
I’ve run more Cassandra than anyone on the planet. Also more FoundationDB. And more Postgres than I can count. Likely 3 orders of magnitude more than you.
1. Any db at that scale requires a team. Thousands of cores is gonna be either paying saas or paying a team and I don’t care what the tech is. You’re gonna fuck up any db on the market letting your app eng run a thousand cores of Postgres or MySQL or couchbase or cockroach or tidb.
2. Cassandra at scale has a demonstrably better cost and availability slo than literally annything else I’ve ever tried to run, as long as you have a competent team. And I had a great team. But we had a great team because we understood that at million core scale it’s a core of the product and you invest in your core. But a good team of a dozen people can run a million cores at 6-8 nines, at lower cost per byte or cost per query than any db-aas you can buy online from any cloud vendor. I know this because we checked, and I’ve been ahead every time.
3. The db isnt fun to run. But it’s getting way better. If you think this is 2009, it’s not. The folks working on it in 2026 know what they’re doing and the last decade has made it measurably safer and easier.
Nothings free, but I’d bet my salary I can run damn near any company on Cassandra cheaper than any of your alternatives. Even all the vendors who keep selling against Cassandra. A competent team on Cassandra is going to beat your vendor product 90% of the time.
What exactly is your point, besides the self aggrandising? None of this is news to anyone, but I'm sure we're all glad to hear you're so shit-hot at everything.
So what is the alternative? It took me a while to get around the data schema / modelling that a database like Cassandra imposed, but when it finally clicked it really clicked for me, so much so that I wish I could have the same performance footprint of consistent hashing / sharding with other databases.
We are using ScyllaDB, but since they discontinued their Open Source version we're stuck on the last supported version, and reading about the operational headaches of Cassandra does not inspire confidence to give it a go.
Drop in with same schematics? It's not going to be easy, I don't have an answer. It's unpopular, but probably I'd architect my software more towards a sharded mongo cluster for write heavy workloads these days, rather than lean on cass/scylla/cockroach.
Mongo bit me a lot pre-wiredtiger and I refused to forgive them for years, but I'm using it again since 8.x and it's really... Boring, Scalable. There are many ways to skin a cat though, and I know not everyone has the luxury of redesigning their apps to fit the persistence layer.
Second that question not only to OP/GP.
Comparing Scylla and Cassandra - the choice is obvious, but current offering is a complete no-go.
Also Mongo is catching up in the use case we have and it’s even more scary. We are all the time at the exact edge of Mongo scalability and I’m talking about very good, few-years-invested-in setup on NMVe that is really delicate…
We live in wildly different worlds, I think. How is postgres comparable to 6-12 nodes of cassandra? One really giant one with 200k worth of storage? Is there even a multi-master setup yet? every time i’ve looked it’s just around the corner. multigres from supabase seems promising, but im not touching it for at least a couple years…
I’d love to use postgres, and have in various times in the past and love it (it never broke) — but one machine isn’t nearly big enough for really any of my current applications.
I have high hopes for Neki from Planetscale but then again if you are not stuck on Postgres for whatever reason, Vitess with MySQL is about as battle tested as it gets.
You can use a distributed database like yugabytedb or crdb. They are horizontally scalable, support postgres (yugabytedb has much better postgres compatibility). Yugabytedb also supports the Cassandra API.
These days if i need to scale sideways — mongo. I shifted most of the work my db was doing into my apps, and it’s… fine? It requires a different architecture tho for sure and has its own problems, but I’ll take those over the pain of cassandra gladly.
its not a drop in. There is no reasonable way of replacing a huge existing cassandra cluster with scylladb. You have to transition as it was a new database.
Directly and indirectly, there are a lot of other knock on effects. Value objects (only recently being added) help. You could treat it similar to an HFT application, having ephemeral workers that don't do GC and dies on heap exhaustion, with a NIO coordinator that tries to avoid all garbage creation. But why at this point, square peg round hole.
I see a few comments talking about the pain cassandra inflicts on ops, and that's fine, I agree with almost all of them. What the comments assume readers to know and understand is the absolute horror of foot guns, backup nightmares, data loss scenarios with seemingly safe choices around quorum and of course the performance purgatory lined with tombstones. Honestly, your workload is not big enough for postgres, trust me. It is like trying to do word count on a 5 TB file with flink, can you? should you though?
Yep, the pain is real. We don’t mention sstableloader in polite company. Add in kubernetes and getting your rf/racdc.properties
right in a 3 az cluster so everything doesn’t just die on a netsplit with local_quorum (harder than it sounds) the joys of the ‘repair’ cronjob, backups, and commitlogs only occasionally not corrupting to the point the 8th out of the 15 nodes in your kube sts can’t restart after a bad exit() (e.g during worker upgrades —- good luck with this one!) …. yeah. hate.
Postgres isn’t the answer for my workload tho we really tried, i even wrote custom sharding for our postgres and it couldn’t handle the writes the cassandra setup shrugs off, but for sure this isn’t a ‘normal’ requirement most applications have.
I'm sure for some workloads, cassandra is 'ideal' or at least was 10 years ago.
What I can say though (as someone running thousands of cpus worth of cassandra at the moment); is that I deeply dislike this database. It's an operational nightmare, and I will not miss it even the slightest bit if it goes away forever.
On the other hand, I'm sure glad to see a bit of life in cassandra dev, it seems to have been dormant for a while; but would I pick it for new applications in 2026? No. I'd strongly recommend against using it unless you have the money to pay for a dedicated team to maintain it, and even then just.. No.
Feels like a thread for me
I’ve run more Cassandra than anyone on the planet. Also more FoundationDB. And more Postgres than I can count. Likely 3 orders of magnitude more than you.
1. Any db at that scale requires a team. Thousands of cores is gonna be either paying saas or paying a team and I don’t care what the tech is. You’re gonna fuck up any db on the market letting your app eng run a thousand cores of Postgres or MySQL or couchbase or cockroach or tidb.
2. Cassandra at scale has a demonstrably better cost and availability slo than literally annything else I’ve ever tried to run, as long as you have a competent team. And I had a great team. But we had a great team because we understood that at million core scale it’s a core of the product and you invest in your core. But a good team of a dozen people can run a million cores at 6-8 nines, at lower cost per byte or cost per query than any db-aas you can buy online from any cloud vendor. I know this because we checked, and I’ve been ahead every time.
3. The db isnt fun to run. But it’s getting way better. If you think this is 2009, it’s not. The folks working on it in 2026 know what they’re doing and the last decade has made it measurably safer and easier.
Nothings free, but I’d bet my salary I can run damn near any company on Cassandra cheaper than any of your alternatives. Even all the vendors who keep selling against Cassandra. A competent team on Cassandra is going to beat your vendor product 90% of the time.
What exactly is your point, besides the self aggrandising? None of this is news to anyone, but I'm sure we're all glad to hear you're so shit-hot at everything.
So what is the alternative? It took me a while to get around the data schema / modelling that a database like Cassandra imposed, but when it finally clicked it really clicked for me, so much so that I wish I could have the same performance footprint of consistent hashing / sharding with other databases.
We are using ScyllaDB, but since they discontinued their Open Source version we're stuck on the last supported version, and reading about the operational headaches of Cassandra does not inspire confidence to give it a go.
Drop in with same schematics? It's not going to be easy, I don't have an answer. It's unpopular, but probably I'd architect my software more towards a sharded mongo cluster for write heavy workloads these days, rather than lean on cass/scylla/cockroach.
Mongo bit me a lot pre-wiredtiger and I refused to forgive them for years, but I'm using it again since 8.x and it's really... Boring, Scalable. There are many ways to skin a cat though, and I know not everyone has the luxury of redesigning their apps to fit the persistence layer.
> We are using ScyllaDB, but since they discontinued their Open Source version we're stuck on the last supported version
Do you think you are missing anything by being on scylla oss?
Second that question not only to OP/GP. Comparing Scylla and Cassandra - the choice is obvious, but current offering is a complete no-go.
Also Mongo is catching up in the use case we have and it’s even more scary. We are all the time at the exact edge of Mongo scalability and I’m talking about very good, few-years-invested-in setup on NMVe that is really delicate…
i'd say 99% of usecases will be served fine with Postgres, or a managed instance via AWS
We live in wildly different worlds, I think. How is postgres comparable to 6-12 nodes of cassandra? One really giant one with 200k worth of storage? Is there even a multi-master setup yet? every time i’ve looked it’s just around the corner. multigres from supabase seems promising, but im not touching it for at least a couple years…
I’d love to use postgres, and have in various times in the past and love it (it never broke) — but one machine isn’t nearly big enough for really any of my current applications.
>multigres from supabase seems promising
I have high hopes for Neki from Planetscale but then again if you are not stuck on Postgres for whatever reason, Vitess with MySQL is about as battle tested as it gets.
You can use a distributed database like yugabytedb or crdb. They are horizontally scalable, support postgres (yugabytedb has much better postgres compatibility). Yugabytedb also supports the Cassandra API.
cocroach is also open source no anymore.
ydb is not clear if easier to maintain at scale.
[flagged]
I also run a large-ish cassandra cluster, and I dont really disagree, but then what else?
I would not recommend starting new projects with it either, but what else will you use?
These days if i need to scale sideways — mongo. I shifted most of the work my db was doing into my apps, and it’s… fine? It requires a different architecture tho for sure and has its own problems, but I’ll take those over the pain of cassandra gladly.
Wasn't ScyllaDB designed to be a drop in? The idea of running a Java or Golang DB sounds like fresh hell to me.
its not a drop in. There is no reasonable way of replacing a huge existing cassandra cluster with scylladb. You have to transition as it was a new database.
Issue is that it’s no longer open source
Why is java bad here...? Because of GC?
Directly and indirectly, there are a lot of other knock on effects. Value objects (only recently being added) help. You could treat it similar to an HFT application, having ephemeral workers that don't do GC and dies on heap exhaustion, with a NIO coordinator that tries to avoid all garbage creation. But why at this point, square peg round hole.
yeah, i suppose one could start with scylladb as their base today.
I see a few comments talking about the pain cassandra inflicts on ops, and that's fine, I agree with almost all of them. What the comments assume readers to know and understand is the absolute horror of foot guns, backup nightmares, data loss scenarios with seemingly safe choices around quorum and of course the performance purgatory lined with tombstones. Honestly, your workload is not big enough for postgres, trust me. It is like trying to do word count on a 5 TB file with flink, can you? should you though?
Yep, the pain is real. We don’t mention sstableloader in polite company. Add in kubernetes and getting your rf/racdc.properties right in a 3 az cluster so everything doesn’t just die on a netsplit with local_quorum (harder than it sounds) the joys of the ‘repair’ cronjob, backups, and commitlogs only occasionally not corrupting to the point the 8th out of the 15 nodes in your kube sts can’t restart after a bad exit() (e.g during worker upgrades —- good luck with this one!) …. yeah. hate.
Postgres isn’t the answer for my workload tho we really tried, i even wrote custom sharding for our postgres and it couldn’t handle the writes the cassandra setup shrugs off, but for sure this isn’t a ‘normal’ requirement most applications have.
Thanks for this, really looking forward to the release of Accord